OCCASIONAL PAPER · 2020. 4. 1. · NIR Net international reserves RR Reserve requirements V ©International Monetary Fund. Not for Redistribution. This page intentionally left blank

OCCASIONAL PAPER

Monetary Policyin Dollarized Economies

By a Staff Team led byTomas J.T. Balino, Adam Bennett, and Eduardo Borensztein

and comprisingAndrew Berg

Zhaohui ChenAlain Ize

David O. RobinsonAbebe Aemro Selassie

Lorena Zamalloa

INTERNATIONAL MONETARY FUNDWashington DC

171

©International Monetary Fund. Not for Redistribution

© 1999 International Monetary Fund

Production: IMF Graphics SectionFigures: In-Ok YoonTypesetting: Alicia Etchebarne-Bourdin and Julio R. Prego

Library of Congress Cataloging-in-Publication Data

Monetary policy in dollarized economies / Tomas Baliiio, Adam Bennett,and Eduardo Borensztein.

p. cm.—(Occasional paper; 171)I S B N 1-55775-757-7 (paper)

1. Monetary policy—Developing countries. 2. Foreign ex-change—Developing countries. 3. Dollar, American. I. Bennett,Adam. II. Borensztein, Eduardo. III. Title. IV. Series: Occasionalpaper (International Monetary Fund); no. 171.

HG1496.B345 1999332.4'91724—dc21 98-43299

CIP

Price: US$18.00(US$15.00 to full-time faculty members and

students at universities and colleges)

Please send orders to:International Monetary Fund, Publication Services

700 19th Street, N.W., Washington, D.C. 20431, U.S.A.Tel.: (202) 623-7430 Telefax: (202) 623-7201

E-mail: publications @imf.orgInternet: http://www.imf.org

recycled paper


http://www.imf.org

Contents

Abbreviations v

Preface vii

I Overview 1

II Trends and Explanations 5

Currency Substitution Versus Asset Substitution 5Foreign Currency Deposits 5Cross-Border Deposits 7Dollar Currency in Circulation 9

III Risks and Benefits of Dollarization 12

Benefits 12Risks 13

IV Monetary and Exchange Rate Policy in a Dollarized Economy 14

Exchange Rate Regime 14Choice of Monetary Target 15

V Operational Issues in Dollarized Economies 17

Implementation of Monetary Policy 17Implications for the Payments System 17Prudential Supervision 21Implications for the Central Bank 23

VI Measures to Affect Dollarization 24

Alternative Financial Instruments 24Policies That Create an Interest Rate Wedge 25Direct Restrictions on Foreign Currency Deposits 26Policies to Encourage the Use of Local Currency Cash 28

VII Program Design in the Presence of Dollarization 29

General Program Design 29Design of Performance Criteria 33Dollarization and Program Performance 37

VIII Conclusions 43

References 44

Box

II 1. Flows of U.S. Dollar Cash to Argentina 10

iii


Tables

I 1. Reported Ratios of Foreign Currency Deposits (FCD) BroadMoney in Countries with IMF Arrangements Since 1986

II 2. Cumulative Net Inflows of U.S. Dollar Currency, 1989-96V 3. Payment and Regulatory Arrangements in Selected

Dollarized Economies4. Reserve Requirements on FCD for Selected Countries

VII 5. Recommendations for the Design of IMF Programs in thePresence of Dollarization

6. Some Features of Dollarized Economies7. Gross Reserves in Months of Imports of Countries Reporting

Information on FCD/Broad Money Ratios8. Selected Indicators of Measures Influencing Dollarization9. Selection of Performance Criteria in Sample Countries

10. Inflation in Countries Reporting Information on FCD/BroadMoney Ratios

11. Velocity in Countries Reporting Information on FCD/BroadMoney Ratios

12. Inflation Target Versus Outcome for the First ProgramYear in IMF Arrangements

13. Money Multiplier

Figures

II 1. Dollarization and Stabilization2. FCD and Cross-Border Deposits

V 3. Bolivia: Dollarization TrendsVI 4. Chile: Indexation and Dollarization

5. Egypt: Asset Substitution and Real Deposit Rates6. Hungary: Interest Rate Differential and Asset Substitution7. Bolivia, Mexico, and Peru: Share of Deposits

VII 8. Turkey: FCD/M3 Ratio and Lira Deposit Rates

210

1822

3031

323436

38

39

4142

68212425262733

The following symbols have been used throughout this paper:

. . . to indicate that data are not available;

— to indicate that the figure is zero or less than half the final digit shown, or that the itemdoes not exist;

- between years or months (for example, 1994-95 or January-June) to indicate the yearsor months covered, including the beginning and ending years or months;

/ between years (for example, 1994/95) to indicate a crop or fiscal (financial) year.

"Billion" means a thousand million.

Minor discrepancies between constituent figures and totals are due to rounding.

The term "country," as used in this paper, does not in all cases refer to a territorial entity thatis a state as understood by international law and practice; the term also covers some territor-ial entities that are not states, but for which statistical data are maintained and provided in-ternationally on a separate and independent basis.

CONTENTS

iv


Abbreviations

CBD Cross-border depositsCMIR Currency and Monetary Instruments ReportsDCC Dollar currency in circulationFCD Foreign currency depositsFCL Foreign currency loansFCR Foreign currency reserves (required for FCD)LCD Local currency depositsNDA Net domestic assetsNFA Net foreign assetsNIR Net international reservesRR Reserve requirements

V


This page intentionally left blank


Preface

Dollarization, which is a common feature in many countries, particularly develop-ing countries and countries in transition, has important economic implications. ThisOccasional Paper analyses the costs and benefits of dollarization and explores its pru-dential and monetary policy implications. In particular, the paper reviews dollariza-tion trends; explores the various monetary and exchange rate policy strategies thatmay be pursued in the presence of dollarization; and examines the implications ofdollarization for the conduct of monetary policy, the issue of prudential norms andregulations, and the design of IMF programs. The paper also considers the effective-ness and drawbacks of measures to limit dollarization.

The authors thank Guillermo Calvo and Peter Garber, who provided comments onearly versions of the paper, as well as many colleagues at the IMF but especially PaulMasson, Ratna Sahay, and Tessa van der Willigen, and members of the Executive Boardfor valuable comments and a stimulating Board seminar discussion in January 1998.

Kiran Sastry and Manzoor Gill provided research assistance, and Magally Bernal,Usha David, Jacqueline Greene, Sylvia Palazzo, and Bogna Jezierska secretarial as-sistance. James McEuen of the External Relations Department edited the paper forpublication and coordinated production.

The views expressed here are the sole responsibility of the authors and do not nec-essarily represent the opinions of the Executive Board of the IMF or other membersof the IMF staff.

vii


This page intentionally left blank


I Overview

D ollarization, the holding by residents of a signif-icant share of their assets in the form of foreign-currency-denominated assets, is a common feature ofdeveloping countries and transition economies and isthereby typical—to a greater or lesser extent—ofmany countries that have IMF-supported adjustmentprograms.1 Of those countries that have had arrange-ments with the IMF at one time or another during thepast ten years, at least half are dollarized, and a signif-icant number are highly dollarized (Table 1).2 Thispaper explores the general question of the costs andbenefits of dollarization for a country's economy. Inaddition, it examines the issues that dollarizationposes for the formulation and conduct of monetarypolicy, as well as for IMF program design.3

The paper focuses on dollarization of the mone-tary sector, and in particular on holdings by residentsof foreign currency deposits (FCD) and, where dataare available, of foreign currency cash. The paperrecognizes, however, that dollarization of monetaryassets often is part of a larger process of financialmarket integration. For example, dollarization in theloan portfolio of banks is an important phenomenon,and the paper touches on this too. Cross-border de-posits (bank deposits of residents in foreign coun-tries) also play an important role as close substitutesfor domestic FCD. The paper explores the variousmonetary policy strategies that may be pursued inthe presence of dollarization, considers the implica-tions of dollarization for the practical applicationand instruments of monetary policy, and examinesthe manner in which dollarization has influenced thedesign of IMF programs. The paper's conclusionsare as follows.

lThe foreign currency is usually, but not always, the U.S. dol-lar. The term "dollarization" serves as a shorthand in this paperfor the use of any foreign currency.

2Of the 99 countries that have had IMF arrangements since1986, 52 reported data on foreign currency deposits (FCD) to theIMF. In those countries reporting data, the median level of FCDover broad money in 1995 was 21.8 percent.

3Several of the issues discussed in this paper (for example, for-eign currency risks) arise also in other foreign exchange opera-tions, even if the economy is not dollarized in the sense discussedabove.

• The benefits of dollarization include closer inte-gration with international markets, exposure tocompetition from these markets, and the avail-ability of a more complete range of assets for do-mestic investors. In countries in which inflation-ary experience has destroyed confidence in thelocal currency, dollarization can sometimes helpto remonetize the economy, restore local inter-mediation, and reverse capital flight. The costsof dollarization include the loss of seignorageand a potential for greater fragility of the bank-ing system. Such fragilities can limit the policyoptions available to the authorities, as well as putan additional burden on the central bank aslender of last resort.

• Dollarization can complicate the choice of inter-mediate targets of monetary policy by introduc-ing a foreign currency component into the moneysupply. The suitability of a target that includes, orexcludes, foreign currency depends on the tar-get's relationship with output and prices, and thisis essentially an empirical matter. It is possible,however, that no reliable aggregate can be found.This problem, which is by no means confined todollarized economies, brings into question thepolicy of monetary targeting as opposed to, forexample, relying on a wider set of indicators. Al-though this issue is beyond the scope of thispaper, there are good reasons to believe thatdollar-denominated assets should play some roleamong the set of relevant indicators for monetarypolicy under any alternative approach.

• While the general considerations regarding thechoice of exchange rate system also apply todollarized economies, the prevalence of cur-rency substitution (the use of foreign-currency-denominated assets for transactions) tends tostrengthen the case for a fixed-rate system. Suchan exchange rate arrangement would protect theeconomy from the effects of potentially exces-sive exchange rate and money market volatility.When attempting stabilization from hyperinfla-tion, in particular, a fixed exchange rate can bean effective instrument in highly dollarized

1


Table 1. Reported Ratios of Foreign Currency Deposits (FCD) Broad Moneyin Countries with IMF Arrangements Since 1986

Country 1990 1991 1992 1993 1994 1995

Highly dollarized economies (FCD/broad money > 30 percent) (18)1

ArgentinaAzerbaijanBelarus2

BoliviaCambodia

Costa RicaCroatiaGeorgiaGuinea-BissauLao P. D. R.

LatviaMozambique3

NicaraguaPeruSao Tome and Principe

TajikistanTurkeyUruguay

MedianAverage

AlbaniaArmeniaBulgariaCzech Republic3

Dominica

EcuadorEgyptEl SalvadorEstoniaGuinea4

HondurasHungaryJamaicaJordanLithuania

Macedonia, FYRMalawiMexico2

Moldova3

Mongolia

Pakistan5

PhilippinesPolandRomaniaRussia

Sierra LeoneSlovak RepublicTrinidad and TobagoUgandaUkraine

34.2......70.8

. . .

...

...

...41.542.0

...

...

.... . .

...

...23.280.1

41.748.6

2.1

...12.0

...

...

...

...

...

...

...

...12.2

...12.5..................

2.617.4

31.4

...

...

...

...

...12.0

...

35.1......

76.8. . .

37.7

...

...34.739.4

...11.828.759.9

...

...29.778.5

36.443.3

37.1......

80.826.3

31.9......

31.636.8

...16.737.465.0

...

...33.776.2

36.843.0

40.414.840.683.938.8

29.553.8

...30.9

41.4

27.223.245.670.2

...

...37.973.3

40.443.4

43.258.954.381.951.8

30.350.280.131.134.4

27.525.348.664.238.3

...45.874.1

48.649.4

43.950.330.782.356.4

31.057.430.831.235.6

31.132.654.564.031.9

33.746.176.1

39.745.5

Moderately dollarized economies (FCD/broad money < 30 percent) (34)1

1.3...

33.4...

3.0

...50.7

1.4

...6.5

3.116.5

13.0.........

3.9......

8.918.0

24.73.9

...

...

...

...10.5

...

23.8...

23.4...

3.9

...37.3

1.023.0

6.9

5.114.321.312.8

...

...

...4.1

...7.5

11.921.024.817.9

...

...

...

...11.5

...

20.4

...20.3

...3.5

2.826.7

0.93.810.0

7.618.719.511.5

...

...

...3.6

...33.0

13.922.628.829.029.5

3.311.56.9

15.719.4

18.541.632.6

7.22.5

5.423.4

0.69.99.4

11.420.428.112.227.0

...10.66.2

10.319.5

13.620.928.522.128.8

7.813.012.613.332.0

...20.428.4

5.91.5

...25.1

1.711.4

9.6

13.026.625.015.225.9

18.18.07.2

11.020.5

. . .21.520.421.720.6

16.511.113.613.526.9

2

I OVERVIEW


Table 1 (concluded)

Country

Uzbekistan2

VietnamYemenZambia

MedianAverage

MemorandumSelected industrial countries

GreeceNetherlandsUnited Kingdom

1990

...

...

...12.112.8

11.58.7

11.4

1991

...

...10.8

...9.7

13.3

13.27.27.7

Sources: IMF, IMF Staff Country Reports and International Financial Statistics (IFS).

1992

20.125.912.1

...14.315.9

14.87.2

10.5

1993

5.120.919.7

...15.715.0

16.63.9

10.9

1Classification based on observations for 1995; countries in bold are those selected for review.2Latest year's observation for March.3Latest year's observation for June.4Latest year's observation for September.5Fiscal year.

1994

22.520.420.7

8.1

13.617.2

15.04.7

12.6

1995

15.519.720.916.2

16.516.4

21.64.4

15.4

economies. The same conclusion does not applywhen dollarization reflects only asset substitu-tion (the holding of foreign-currency-denomi-nated assets as stores of value).

• Dollarization requires the adoption of specialprudential measures. The banking system mustbe able to withstand significant exchange rateadjustments, as well as possibly larger-than-normal swings in capital flows. To deal with thelatter, commercial banks or the central bank needto hold a larger-than-normal volume of interna-tional reserves, or to arrange external lines ofcredit. Limits to banks' foreign exposure posi-tions need to be monitored carefully, as do off-balance-sheet operations that could entail for-eign exchange risk. Since devaluations cannotshrink the value of dollar claims, steps have tobe taken to ensure that banks do not incur unduerisks in lending to dollar borrowers that do nothave the capacity to honor their obligationswhen devaluations occur.

• Should dollarization be discouraged? The an-swer depends on the role of dollarization in theeconomy. Asset substitution may be a natural ac-companiment of the opening of financial mar-kets, and in this respect it should be welcome.Moreover, globalization of financial marketswill likely lead to some dollarization.4 Nonethe-

4Other special circumstances may lead to dollarization. For in-stance, European countries that will not initially participate in theeuro arrangements but that trade heavily with euro members arelikely to see part of the demand for money shift to euros. In addi-

less, as with other forms of capital market liber-alization, the proper sequencing of policies is es-sential. Asset substitution—and more especially,currency substitution—may also reflect the ab-sence of macroeconomic stability and the exis-tence of distortions in financial markets. In thesecircumstances, dollarization may complicate sta-bilization and cause additional volatility. How-ever, in circumstances where it becomes verydifficult to reestablish quickly stability of the na-tional currency, economic well-being wouldlikely be reduced by any administrative mea-sures to reduce dollarization, and a case might bemade for acceptance of continued dollarizationin the quest for stability.

• Macroeconomic stability is the first priority indealing with dollarization, but this may not, inand of itself, be sufficient to reverse it. Othermeasures, such as the liberalization of domesticinterest rates, the establishment of a competi-tive domestic currency payments system, andthe development of domestic financial instru-ments are also steps that can help "dedollarize"an economy. More direct measures to reversedollarization, however, can be problematic.Regulatory limits on FCD or punitive reserverequirements on dollar deposits may simplydrive dollars offshore, while forced conversionswill undermine confidence and may also en-courage capital flight.

tion, very small economies in which tourism from abroad is animportant sector are also likely to be dollarized.

Overview

3


• In general, IMF programs have treated dollariza-tion as a symptom to be lived with rather than di-rectly attacked and have concentrated on macro-economic stabilization, in common with programsfor nondollarized economies. In fact, dollarizedcountries with IMF programs have done onlyslightly worse in terms of meeting inflation targetsthan nondollarized economies with programs, al-though there is evidence of higher volatility in theformer group. In some cases, concerns have arisen

regarding the soundness of the banking system indollarized economies. In other cases, the conven-tional practice of treating required reserves onFCD as a domestic, rather than foreign, liabilitymay have put program objectives at risk. Morebroadly, program design in the presence of dollar-ization requires a more thorough analysis thanusual in the selection of intermediate monetarytargets and of the relationship of dollarization withultimate targets.

I OVERVIEW

4


II Trends and Explanations

I n general terms, dollarization is a response to eco-nomic instability and high inflation, and to the de-sire of domestic residents to diversify their assetportfolios. In conditions of hyperinflation, dollariza-tion is typically quite widespread because the publicseeks protection from the cost of holding assets de-nominated in domestic currency. But, remarkably,the increase in dollarization in some Latin Americanand Asian countries has continued and even acceler-ated in recent years following successful stabiliza-tion. Some authors have pointed to ratchet effects incurrency substitution to explain this development.But there is suggestive data indicating that the in-crease in local holdings of dollar assets in the 1990salso resulted from the reversal of capital flight andremonetization.

Currency Substitution VersusAsset Substitution

To understand these developments, it is useful todistinguish between two motives for the demand forforeign currency assets: currency substitution andasset substitution.5 Currency substitution occurswhen foreign-currency-denominated assets are usedas a means of payment, while asset substitution oc-curs when foreign-currency-denominated assetsserve as financial assets (store of value) but not as ameans of payment or unit of account. Currency sub-stitution typically arises under conditions of high in-flation or hyperinflation when the high cost of usingdomestic currency for transactions prompts the pub-lic to look for alternatives. Asset substitution resultsfrom the public's allocation decisions in view of therisk and return characteristics of domestic and for-eign assets. Historically, foreign-currency-denomi-nated assets have provided the opportunity of insur-ing against major macroeconomic risks in manydeveloping countries. Where indexed monetary as-

sets have been available, as in Chile and Brazil, dol-larization has been much less widespread, suggest-ing that indexation may be an alternative to dollar-ization in the face of macroeconomic instability.Even in countries that currently enjoy stability, for-eign-currency-denominated assets may still provideinsurance against the probability, small though itmay be, of a return to inflation and devaluation, andthey may also contribute to a reduction in overallrisk in a balanced portfolio.6

Because of their close connection with monetarydevelopments, this section focuses on three types ofdollar-denominated financial assets: FCD in the do-mestic banking system, dollar currency in circula-tion within the domestic economy (DCC), and cross-border deposits held at banks abroad (CBD). Thelast of these may not be considered representative ofdollarization because it is located abroad, but it ishighly relevant because of its close substitutabilitywith FCD. Because reliable information is availableonly for FCD, most studies are based on this mea-sure only.7 But even the incomplete informationavailable on the other assets can shed some light onrecent developments.8

Foreign Currency Deposits

FCD constitute a significant share of broad moneyin several developing countries and transitioneconomies that have made use of IMF resources inrecent years, reaching over 50 percent in Azerbaijan,Bolivia, Cambodia, Croatia, Nicaragua, Peru, andUruguay in 1995 (see Table 1). An FCD ratio of

5This distinction is standard in the literature. See the useful sur-veys by Calvo and Vegh (1996) and Giovannini and Turtelboom(1994). McKinnon (1996) terms the two motives direct currencysubstitution and indirect currency substitution.

6While the discussion in this paper focuses chiefly on the assetsof the public, dollarization of deposits has typically been accom-panied by dollarization of lending.

7Previous empirical studies on dollarization conducted in theIMF include those by Agenor and Khan (1996), Clements andSchwartz (1993), El-Erian (1988), Mueller (1994), Sahay andVegh (1996), Savastano (1992), and Savastano (1996).

8Dollarization as asset substitution is also associated with othernonmonetary assets, such as foreign-currency-denominatedstocks and bonds and borrowing in foreign currency. But theseassets are more removed from monetary developments and areconsidered more tangentially in this study.

5


II TRENDS AND EXPLANATIONS

Figure 1. Dollarization and Stabilization1

(Ratio of foreign currency deposits, FCD, to broad money, M3; in percent)

around 15-20 percent appears common in countrieswhere residents are allowed to maintain foreign cur-rency accounts. On account of their large size, per-sistence, and volatility over time, FCD develop-ments in two regions—Latin America and thetransition economies of Eastern Europe and the for-mer Soviet Union—are particularly worth highlight-ing. Figure 1 displays the ratio of FCD to broadmoney (inclusive of FCD) for a selected group ofcountries in these two regions.9

With the advent of market reforms in the transi-tion economies during the early 1990s, restrictionson FCD were generally eased (Sahay and Vegh,1996). As a result, the FCD ratio rose rapidly, reach-ing peak levels of 30-60 percent in most transition

9In Africa, (measured) dollarization is less widespread than inother areas because FCD are not allowed or are severely re-stricted in most cases. However, there are indications of extensivedollarization in banknotes.

economies during the 1990-95 period (see Table 1).High inflation rates, negative real interest rates ondomestic-currency-denominated assets, and sharpdevaluations that increased the domestic currencyvalue of dollar deposits contributed to the rise in theFCD ratio. Following price stabilization, the FCDratio declined sharply in countries such as Armenia,Estonia, Poland, and Mongolia. The valuation effectof substantial real appreciation also contributed tothis decline, since it more than offset the rise in thedollar volume of FCD in several cases.

Abstracting from the periods when the convert-ibility of FCD was suspended, the ratio of FCD tobroad money (inclusive of FCD) increased steadilyin several Latin American countries during the1970s and the 1980s, in a period of high inflationand a gradual easing of restrictions on FCD (Savas-tano, 1996). The 1990s witnessed a further easing ofFCD restrictions in most countries, and FCD ratioscontinued to rise despite a sharp deceleration in in-

6


Cross-Border Deposits

Figure 1 (concluded)

Sources: National authorities; and IMF staff estimates.1Vertical line denotes end of period of high inflation. For Uruguay, the gradual nature of the stabilization

process does not permit one to identify a precise timing for stabilization.

flation rates, reaching 80 percent in Bolivia andUruguay, over 60 percent in Peru, and nearly 50 per-cent in Argentina by 1995-96 (see Figure 1).

Although the history of high inflation and financialinstability is closely linked to the prevalence of dol-larization in several Latin American countries, it isremarkable that the FCD ratio increased sharply aftercountries successfully battled inflation in the late1980s and early 1990s. (The approximate time of sta-bilization is shown by the vertical line in Figure 1.)One possible explanation is "hysteresis" or someform of nonreversibility in the process of dollariza-tion (see Guidotti and Rodriguez, 1992). This couldemerge, for example, because changing uses andpractices regarding the settlement of transactions is aslow process that involves (informal) institutionalchanges and takes place only when there are signifi-cant benefits to be gained by switching currencies.

For the same reasons, a reversal of dollarization afterstabilization would also be slow, especially if thereare no significant benefits to be gained from switch-ing back to the domestic currency as a means of pay-ment.10 It is also possible that financial innovationand liberalization have permanently reduced the costof holding assets denominated in dollars.

Cross-Border Deposits

While the hysteresis argument may explain the per-sistence of dollarization, it would still not account forthe cases of steady increase in dollarization after stabi-

10Kamin and Ericsson (1993) estimated a money demand func-tion with "ratchet" effects for Argentina for a period that includesa hyperinflation episode.

7



Figure 2. FCD and Cross-Border Deposits (CBD)(In millions of U.S. dollars unless otherwise noted)

lization. The surge in capital inflows to developingcountries in the 1990s offers an additional, or perhapsalternative, explanation for the persistent growth in theFCD ratio in the poststabilization period in LatinAmerican countries.11 It is quite suggestive that the in-crease in FCD (in dollar terms) coincided with the de-crease in CBD in the 1990s for various Latin Ameri-can countries, including the short-lived reversal in thecases of Mexico and Argentina at the time of the Mex-ican peso crisis (Figure 2). This suggests that the in-crease in FCD in part reflected a shift in residents'portfolios from CBD to dollar deposits in the domesticbanking system. In this sense, the increase in domesticdollarization would reflect an increase in confidencein the domestic economy (although not necessarily inthe domestic currency), rather than a persistent lack of

credibility.12 This could be part of the process of gen-eral remonetization of the economy, as well as an in-crease in dollar-denominated lending by domesticbanks. Moreover, because the persistent increase inthe FCD ratio seems to be related to shifts from CBDto FCD, it is possible that the increase in the volumeof FCD does not represent an increased volume ofoverall dollar assets. If a more comprehensive mea-sure were available, one inclusive of cash and cross-

11Calvo and Vegh (1992, 1996) and Savastano (1992, 1996)also mention capital flows as a driving factor for dollarization.

12While it is true that, according to available data, the declinein CBD is less than the increase in FCD in absolute terms, itshould be noted that the CBD figures are likely to underestimatethe actual stock of CBD. For legal and tax-related reasons, resi-dents of one country may wish to transfer assets to a companybased in a third country, which appears to be the holder of the de-posit. (In fact, the three top countries of residence of nonbank de-positors in the Western Hemisphere are Cayman Islands, Panama,and the Netherlands Antilles.) This means that the actual declinein CBD was probably larger than reported.

8


Dollar Currency in Circulation

Figure 2 (concluded)

Sources: Bank for International Settlements (BIS); national authorities; and IMF staff estimates.

border deposits, the share of this broader set of foreigncurrency assets might even be seen to be declining.


Although estimates of the amount of DCC in thehome country do not exist, there is some suggestivedata on flows of U.S. currency to other countries,collected from U.S. Customs forms called Currencyand Monetary Instruments Reports (CMIR), whichare completed when currency in amounts of $10,000or higher is physically transported into or out of theUnited States.13 The data displayed in Table 2 show

13These data have been kindly made available by the U.S. Trea-sury Department (Financial Crime Enforcement Network); this isthe first time this information has been released with the exceptionof the data for Argentina used by Kamin and Ericsson (1993).

strikingly large flows of U.S. currency to several de-veloping countries. The cumulation of net flows intothese countries since 1989 reaches or exceeds 12 per-cent of GDP for Argentina (Box 1), Bolivia, Latvia,Russia, and Uruguay and exceeds DCC by a factor oftwo to three. One obvious shortcoming of these datais that much of the currency shipped to some coun-tries ends up in third countries (or returns to theUnited States without being recorded) via capital out-flows, tourism expenditures, and so on. In Russia, forexample, "shuttle" trade (imports by individuals whocross borders to purchase goods in other countries) isconducted mainly in cash. In Argentina, significantoutflows of dollar currency have also apparentlytaken place, including a large volume related to inter-national tourism (Kamin and Ericsson, 1993).

The experience of Latin American and transitioneconomies suggests several stylized facts. First, asrestrictions are lifted, the FCD ratio can be expected

9



Table 2. Cumulative Net Inflows of U.S. Dollar Currency, 1989-96

Country

ArgentinaBoliviaBrazilEstoniaHungary

LatviaPeruPolandRomaniaRussia

TurkeyUkraineUruguayVenezuela

In Millions ofU.S. Dollars

34,737831

1,8904323

82263660944

43,772

3,64837

2,1041,191

In U.S. Dollarsper Capita

1,014103

1228

2

31627162

294

601

66254

As Percent ofLocal Currencyin Circulation

294290

1712

1

17351

73

246

1102

442141

As Percentof GDP

12.411.90.31.20.1

18.41.10.50.1

12.2

2.80.0

11.81.8

As Percent ofForeign Currency

Deposits

147.836.2

...

...

...

...10.47.3

...382.4

17.8

...26.5

...

Sources: Data on cumulative net dollar currency flows from the United States from 1989 through 1996 from Financial Crimes Enforcement Network,U.S. Department of the Treasury, as recorded in the Customs Service Currency and Monetary Instruments Reports (CMIR) forms. See text for discus-sion. Other data from IMF, IFS, and from national authorities (central bank bulletins).

Box 1. Flows of U.S. Dollar Cash to Argentina

Despite shortcomings, the data from U.S. CustomsCurrency and Monetary Instruments Reports (CMIR)can be informative, as the depicted evolution of net dol-lar cash inflows to Argentina suggests. At times of fi-nancial insecurity there are sharp increases in the ship-ment of U.S. dollar currency into the country. This isnoticeable in the two spikes in the first figure. The firstspike took place at the time of the Mexican crisis (end-1994/early 1995), and the second one at the time of theresignation of Minister Cavallo in July 1996. The sec-ond figure shows that the increase in DCC (obtained

Argentina: Net Inflow of Dollar Cashper Quarter(In billions of U.S. dollars)

from cumulative flows) parallels that of FCD, except fora movement in the opposite direction at the time of theMexican crisis (1994—95), which presumably reflects arun on domestic dollar deposits. This evidence suggeststhat DCC may function partly as CBD (as suggested bythe flight from FCD to CBD and DCC during the Mexi-can crisis) and partly as a complement to FCD (as sug-gested by the general growing trend in both aggregates).While this information does not permit one to constructfirm estimates, it is suggestive of the magnitude and rel-evance of the stock of DCC in dollarized economies.

Argentina: Stocks of Foreign CurrencyDeposits, Dollar Cash, and Cross-BorderDeposits per Quarter(In billions of U.S. dollars)

10



to increase as the public takes advantage of the op-portunity to diversify its asset portfolio or to repatri-ate funds that were held in other countries. Whererestrictions are suddenly removed, as in some transi-tion economies, a distinct stock adjustment can beobserved. Second, the existence of macroeconomicimbalances reflected in high inflation rates can leadto currency substitution, especially when other infla-tion-protected liquid financial assets are not avail-

able and more so in situations of hyperinflation.Third, currency substitution can persist even aftercountries have successfully brought down inflationrates, especially when they previously experiencedlong periods of inflation. Finally, even in the ab-sence of currency substitution, dollarization in theform of asset substitution can be a permanent fea-ture, especially as the process of financial deepeningand globalization advances.

I I


Ill Risks and Benefits of Dollarization

I n a fully liberalized and convertible financial sys-tem, savers would prefer to hold a portion of theirportfolio in foreign-currency-denominated assets,simply in order to achieve a desired distribution ofrisk and returns. But such diversification may in-crease systemic risks when financial systems are stillimmature and subject to many distortions. There-fore, dollarization of financial systems in developingcountries presents both advantages and risks. On theone hand, allowing FCD in the domestic financialsystem provides the opportunity to allow greater do-mestic intermediation. On the other hand, because ofcurrency risk and potential balance of paymentsproblems, the systemic risks are high, and dollariza-tion could increase the potential for financial andbanking crises. These considerations imply that thespeed with which it would be advisable to advancetoward a fully convertible financial system and theproper timing for the liberalization of FCD and loanswill be determined by the conditions prevailing inspecific cases.

Benefits

Allowing FCD in the domestic financial systemenhances the opportunity for reintermediation ineconomies that have undergone periods of very highinflation and unstable macroeconomic conditions,during which agents may have become reluctant tohold deposits in the banking system. With a restora-tion of stability, it is likely that confidence will be re-built only gradually. The availability of FCD canspeed up this process to the extent that agents may bemore willing to return to domestic intermediaries ifthey can hold dollar-denominated assets.

Allowing FCD also promotes financial deepening.Domestic banks can expand their operations rapidlyby competing for the FCD held by residents in cross-border accounts. In fact, the growth of domestic bankswould be somewhat limited if they were not allowedto offer FCD, since residents are unlikely to reducesignificantly the share of their savings held in theform of dollar financial assets for some time. In addi-tion, the existence of dollar accounts in domestic

banks could facilitate the integration of the domesticmarket into the rest of the world and lower the cost ofinternational financial transactions. Also, in case offear of devaluation, the availability of FCD may en-courage depositors to shift at least part of their localcurrency deposits (LCD) into FCD rather than CBD.14

Allowing residents to hold FCD can increasecredibility by raising the cost of monetary indisci-pline, thus committing the government to strongerfinancial policies. Similar arguments have beenmade to support the introduction of indexed govern-ment debt. A similar type of commitment strategy isto issue dollar-denominated debt to domestic resi-dents as a substitute for domestic currency debt.However, it is not always the case that "raising thestakes" in this form will guarantee sound policies oravoid crises, and the costs of a crisis would behigher if one did occur.15 Furthermore, when confi-dence is very weak, it would appear that a differenttype of monetary framework, such as a currencyboard, would be more effective, both to bolster fi-nancial discipline and to reinforce confidence. Wheneven this is not sufficient to strengthen the govern-ment's financial discipline sufficiently, full dollar-ization—abolition of the domestic currency alto-gether—could indeed be the only alternative toachieve price stability. This would entail certainlosses for the country: forgoing seignorage, limita-tions to the lender of last resort function, and the lossof the exchange rate instrument. It is true, however,that the value of these resources and instruments islimited in economies that are overwhelmingly dol-larized.16 Yet full dollarization, being difficult to re-

14For a discussion of this issue for the case of Argentina, seeGarcia-Herrero (1997).

15The issuance by the Mexican government of the dollar-indexed tesobonos in 1994 signaled a commitment to a fixed ex-change rate and raised the risks involved in a potential reversal ofcapital flows. On balance, it does not appear that tesobonos had apositive effect on credibility. As suggested by Drazen and Masson(1994), one has to distinguish between credibility of policy makersand credibility of policies, and the latter was very low in this case.

16The exchange rate instrument could still be effective to in-duce changes in relative prices if domestic prices are not rigid indollar terms; that is, if the domestic currency is the predominantunit of account in the economy.

12


verse, effectively precludes the possibility of a re-covery in the demand for domestic money.

Risks

On the negative side, a rapid development of dol-lar-denominated operations in the banking systemincreases the risk of crisis in financial and foreignexchange markets. Capital inflows intermediated bythe banking system may expand gross official inter-national reserves, with a parallel increase in domes-tic short-run liabilities in the form of the increase inbanks' required reserves with the monetary authori-ties. Unless the monetary authorities hold all the re-sultant increase in required reserves in foreign ex-change reserves, there will be a deterioration in theirnet foreign-currency-denominated position. More-over, the rapid expansion in dollar deposits andloans that may result from a return of flight capital islikely to increase the riskiness of the loan portfolioof the domestic banks. Although this increase in riskwould also develop in the domestic-currency loanportfolio, an important difference is that the expan-sion in loans and deposits under the fractional re-serve banking system implies that the total volumeof dollar-denominated assets and liabilities willgreatly exceed the volume of net dollar assets held inthe economy. This lessens the central bank's abilityto act effectively as lender of last resort and in-creases the vulnerability of the banking system tocapital outflows.17 Maturity mismatches betweenbank assets and liabilities in foreign currency wouldincrease banks' vulnerability to volatile capitalflows. For example, withdrawal of short-term, dollarcredits from banks forced the Mexican authorities toprovide substantial dollar loans to the banking sys-

tem in early 1995.18 Further, in the case of a devalu-ation, loan defaults would increase, and the financialposition of banks deteriorate, unless dollar lending islargely to debtors whose net financial position bene-fits from a devaluation, for example exporters. Infact, the central bank may attempt to avoid signifi-cant devaluations because of the likely negative ef-fect on the quality of the banks' loan portfolios, evenif there were otherwise valid macroeconomic rea-sons to devalue.

Dollarization also implies the loss of seignoragerevenues for the monetary authorities (see Fischer,1982). Other things equal, the use of foreign cur-rency reduces the demand for domestic money andimplies a lower level of seignorage for the govern-ment. While the central bank may earn someseignorage from below-market remuneration ofbanks' required reserves in foreign currency, it can-not avoid seignorage losses that result from foreigncurrency in circulation. The data on flows of U.S.dollar currency to various dollarized economies (seeTable 2) suggest that the circulation of foreign cur-rency, and thus the potential seignorage losses, is in-deed substantial in some of the more dollarizedeconomies. For example, if the ratio of U.S. dollarcurrency to GDP is 10 percent, and monetary aggre-gates grow at an average of 15 percent a year,seignorage losses would amount to 1.5 percent ofGDP a year.19

17Moreover, unlike with loans in local currency, the burdenof foreign currency loans (FCL) could not be reduced throughdevaluation.

18As noted earlier, the problems described in this paragraphwould also apply to banking systems exposed to foreign exchangerisk because they are heavily engaged in foreign exchange transac-tions. However, such exposure is likely to be more pervasive incountries in which there is extensive currency or asset substitution.

19The magnitude of the seignorage revenues obtained by theUnited States from the holding of its currency in foreign countriesis relatively small. It is estimated that $200 to $250 billion circu-lated outside the United States in 1995 (Porter and Judson, 1996).If there was no such foreign demand for the U.S. currency, andthe United States had to issue an equivalent amount of short-termtreasury bills, the annual interest cost would be some $10 billionto $15 billion, less than 1/4 of 1 percent of GDP.

13


IV Monetary and Exchange Rate Policyin a Dollarized Economy

The phenomenon of dollarization poses a chal-lenge to the pursuit of a coherent and indepen-dent monetary policy. In responding to this chal-lenge, the authorities must address two key questionsabout the conduct of monetary policy. What is themost appropriate exchange rate/monetary policyregime? What monetary target should the authoritiespursue (in particular, should dollar-denominated as-sets be included in an intermediate target)? This sec-tion addresses these two questions in turn. To orga-nize ideas, it is useful to discuss the cases of currencysubstitution and asset substitution separately, whilebearing in mind that in practice the two may coexist.Currency substitution would in general be accompa-nied by asset substitution, since it is natural to holdfinancial assets denominated in the transaction cur-rency. In contrast, asset substitution does not neces-sarily involve currency substitution.

Exchange Rate Regime

The theoretical literature suggests that a key im-plication of currency substitution is that the volatil-ity of a floating exchange rate will tend to be greater.First, there may be frequent and unexpected shifts inthe use of domestic and foreign money for transac-tion purposes. Because the two currencies serve es-sentially the same purpose, the public may shift be-tween them for a number of not easily identifiablereasons.20 Second, domestic money demand (the de-mand for the domestic component of the monetaryaggregate) will be more sensitive to changes in itsexpected opportunity cost. In addition to the usualeffect of interest rates on overall money demand, thedomestic component of money will also be affectedby changes in its opportunity cost relative to foreignmoney. In other words, the interest elasticity of do-mestic money demand will be higher when currencysubstitution is significant. While the implications of

this higher interest elasticity depend in a complexway on the structure of the economy, one importanteffect is that in a floating-rate system, the exchangerate will be more sensitive to expected changes inthe domestic money supply and other variables thataffect the money market. The higher elasticity mayalso have the opposite effect, however, for differenttypes of shocks. For example, a given random shockto current money demand will have smaller overalleffects, since smaller changes in the exchange rateand interest rates would be required to bring themoney market back into equilibrium.

A clear case for fixing the exchange rate in ahighly dollarized economy is when stabilizing fromvery high inflation or hyperinflation. Currency sub-stitution is likely to be important, and monetaryshocks are likely to predominate, especially becausesuccessful stabilization may result in a large but un-predictable increase in demand for domestic cur-rency. Moreover, in a hyperinflation, foreign cur-rency may assume the role of unit of account, andthe exchange rate may also serve as an approximatemeasure of the price level, which would make therole of fixing the exchange rate very powerful inguiding expectations toward a low-inflation equilib-rium. The stabilization from hyperinflation in Ar-gentina in 1991 and in Croatia in 1994 are exampleswhere an exchange rate anchor helped to stop infla-tion in its tracks within the context of extensive cur-rency substitution.21

Dollarization in the sense of asset substitutionalone does not directly affect a narrow definition ofmoney demand. The availability of dollar deposits indomestic banks, however, has several implicationsfor monetary policy. The most important for thechoice of exchange rate regimes may be that theavailability of dollar deposits in domestic banks also

20For example, in several transition economies, a decline in theacceptability of $100 bills (owing to fears that they would be con-fiscated during the conversion to the new notes) appears to havebriefly raised the demand for domestic currency.

21This conclusion applies not only to exchange rate systems inwhich the rate is formally pegged but also to exchange-rate-basedstabilizations more generally; that is, situations in which macro-economic policies (monetary policy in particular) are geared tomaintaining a stable exchange rate continuously, even though theexchange arrangement is flexible. In Croatia, the anchor was notan explicit, formal one, but ex post this episode can be consideredas an instance of exchange-rate-based stabilization.

14


Choice of Monetary Target

serves to increase capital mobility, since the publiccan potentially shift between dollar-denominated de-posits held with domestic banks and abroad, and be-tween dollar- and domestic-currency-denominateddeposits held with domestic banks. These differentassets are likely to be close substitutes from thepoint of view of savers, and this in turn strengthensthe links between interest rates in dollar deposits athome, international dollar interest rates, and domes-tic currency interest rates. This would limit the con-trol that the central bank can exert on monetary con-ditions, such as the level of interest rates ondomestic currency. In this respect, a flexible ex-change arrangement may be a useful device to in-crease monetary autonomy, which somewhat contra-dicts the general recommendation in the currencysubstitution case.

It is in principle possible to make a judgmentabout the extent to which dollarization representscurrency substitution rather than asset substitution.Countries with a history of hyperinflation are primecandidates for currency substitution, even if theycurrently enjoy price stability. In general, when cur-rency substitution is widespread there is anecdotalevidence of its existence. Large DCC would be an apriori indicator of currency substitution, as would apredominance of demand deposits among the dollar-denominated deposits or a well-developed dollarmoney market. However, in cases where dollar de-posits are not legal, the banking system is not trustedor is underdeveloped (as in Indochina), or where thefear of confiscation is high, foreign currency may bedemanded as a safe asset rather than for currencysubstitution reasons.

Choice of Monetary Target

In a floating exchange rate regime, or a fixed ex-change rate regime with limited capital mobility,dollarization can affect the choice of intermediatetargets of monetary policy. On the view that moneyis targeted because it determines the price levelthrough transaction demand for money, currencysubstitution implies that dollar monetary assets arepart of the relevant concept of money, whereas assetsubstitution implies that they are not.

The empirical literature has shed little light on thedistinction between currency substitution and assetsubstitution in this context. Most work has implicitlyor explicitly assumed away asset substitution in test-ing for currency substitution.22 The traditional ap-proach has been to attempt to identify currency sub-stitution from the coefficients on the rate of returnvariables included in money demand functions.

Specifically, studies added a variable measuring ex-pected exchange rate depreciation to the usual deter-minants of domestic money demand and interpretedthis variable as measuring the opportunity cost ofholding domestic versus foreign currency.23 As Cud-dington (1983) pointed out, however, domesticmoney demand will depend on the rate of exchangerate depreciation even in the absence of currencysubstitution, because the rate of depreciation affectsthe yield of foreign assets, which is an opportunitycost to domestic money. Thus, a test to distinguishbetween currency and asset substitution would in-clude both the rate of return on foreign bonds in do-mestic currency and the rate of depreciation itself inthe money demand regression, with a negative andsignificant coefficient on the rate of depreciationvariable suggesting currency substitution as distinctfrom asset substitution. Unfortunately, these tworates of return variables are closely correlated, par-ticularly in countries likely to have currency substi-tution, and their independent effects are essentiallyimpossible to distinguish.

In this light, a potentially more fruitful approachto identifying currency substitution would start notwith money demand but with the determinants of in-flation. Although money demand functions lookquite similar to asset demand functions, and the ex-planatory variables that may distinguish between thetwo are highly correlated, it is the stock of money,but presumably not of assets, that is closely corre-lated with the volume of transactions and the rate ofinflation. From this point of view, the relevant test ofcurrency substitution is whether foreign monetaryassets belong in the monetary aggregate that predictsinflation in the most reliable way.

Reasoning along these lines, Berg, Borensztein,and Chen (1997) used a vector autoregressionmethodology to examine the strength of the relation-ship between inflation and lagged changes in variousdefinitions of money in Peru.24 They found evidencesuggesting that aggregates that include FCD im-prove inflation prediction in this bivariate frame-work. Although broad, purely domestic currency ag-gregates performed worse than narrow domesticcurrency aggregates, broad aggregates that includeFCD did better. Moreover, they found that an evenbroader aggregate, one that also includes CBD, isthe best predictor of inflation. These results suggestthat aggregates including dollar deposits might beappropriate targets in Peru. However, there is little

22See Savastano (1996) for a useful review of this literature.

23See, for example, Miles (1978) and Bordo and Choudri(1982).

24A number of studies, including Estrella and Mishkin (1996),Friedman and Kuttner (1996), and Feldstein and Stock (1994),have used this type of technique to approach the analogous ques-tion of which monetary aggregate is the most appropriate focus oftargeting.

15


IV MONETARY AND EXCHANGE RATE POLICY IN A DOLLARIZED ECONOMY

reason to expect that this conclusion, illustrative as itis even for Peru, will be the same in each economyor that it will be the same over time, implying a needfor country-specific investigation.25

The choice of a target monetary aggregate in dol-larized economies is, therefore, essentially an empir-ical matter because it is not possible to deduce a pri-ori the asset composition of money demand. Thus,the design of a monetary program would require anempirical investigation of the question whether themonetary aggregate that is most closely associatedwith the final objective—say the inflation rate—includes foreign-currency-denominated assets. Tar-geting the "wrong" aggregate—for example, onethat excludes FCD when in fact the more stable rela-tionship is with one that includes it—would be simi-lar to targeting any monetary aggregate that has aloose or unstable relationship to final targets. The in-termediate target variable will carry information,and targeting it will be less effective. Frequent revi-sions to intermediate target values would become

25The finding for Peru may be affected by institutional changesthat took place during the sample period. The Central Bank ofPeru has found a narrow, purely domestic currency aggregate tobe the most useful intermediate target.

necessary, reducing the usefulness of the monetarytargeting strategy.

The choice of a monetary aggregate as intermedi-ate target for policy should be viewed, however, fromthe broader perspective of finding an aggregate thatprovides useful summary information on monetaryconditions, rather than a target to be strictly pursuedindependent of the behavior of interest rates, the ex-change rate, or other indicators.26 While the questionof the usefulness of money targeting is beyond thescope of this paper, it is worth remembering that theproblem of selecting the appropriate monetary targetis by no means exclusive to dollarized economies. Ithas proven difficult even in comparatively stable in-dustrial countries to rely on targets for monetary ag-gregates in the conduct of monetary policy, and sincethe 1980s monetary targeting has become less andless common even in these countries. In this context,it is still to be expected that dollar deposits wouldplay some role within the set of indicators that thecentral bank would need to watch in assessing mone-tary conditions.

26It may also be difficult to implement an intermediate targetthat includes foreign assets. The lack of reliable data on DCCwould make it difficult to do so.

16


V Operational Issues in DollarizedEconomies

This section first reviews how monetary policyimplementation and payments system arrange-ments have been adjusted in light of heavy dollariza-tion and then discusses the implications of dollariza-tion for prudential regulation and effective banksupervision, and the central bank's role of lender oflast resort.

Implementation of Monetary Policy

Monetary operations can be conducted in eitherdomestic or foreign currency instruments. Althoughconducting all monetary operations in local currencyis simpler in principle and may be used to signal thecentral bank's support for its own currency, the cur-rency of denomination of monetary instrumentsshould be consistent with the main money marketsin the economy, as well as with the authorities' mon-etary targets. Thus, at high levels of dollarization,where the local currency money market is thin, as inBolivia, using dollar instruments is likely to be lesscostly and more effective in implementing monetarypolicy.27 In practice, however, there is considerablevariation in the choice of single or multicurrencymonetary operations in dollarized economies: thetransition countries and Peru, for example, use a sin-gle currency (their own), whereas other countries,such as Argentina, Bolivia, and Uruguay, use bothlocal and foreign currency. Further, the degree ofsubstitutability between dollar-denominated govern-ment bonds and dollar assets available outside thehome country affects the effectiveness of foreign ex-change monetary intervention. The higher is the de-gree of substitutability, the lower is the effectivenessof this instrument.28

27For similar reasons (a high degree of indexation), the CentralBank of Chile's main monetary instrument is an indexed instrument.

28For example, preliminary evidence suggests that the short-runoffset coefficient for Uruguay is higher than that for Bolivia,which implies that foreign exchange monetary intervention wouldbe a more effective instrument in Bolivia than in Uruguay. Offsetcoefficients measure the extent to which changes in the net foreignassets of the central bank resulting from open market operationsare offset by market-originated opposite changes in those assets.

The design of reserve requirements in a dollarizedeconomy is a complex issue. The authorities need toset several interlinked parameters: the level, the re-muneration, and the currency of denomination. Howthese are set affects the cost for banks in fundingthemselves in domestic vis-a-vis foreign currencyand also has prudential implications; insofar as dol-larization reflects currency substitution, the level ofthe requirement will also have monetary implica-tions. The weights to be given to these considera-tions will differ in different situations, and this willaffect the policy advice to be given.29

In heavily dollarized economies, foreign currencyreserve requirements on FCD can play a useful roleas automatic liquidity stabilizers (as in Bolivia, Peru,and Uruguay). When denominated in foreign cur-rency, averaging reserve requirements over the hold-ing period allows banks to draw flexibly on their re-serves, limiting the need for central bank interventionin the dollar money market.30 Reserve requirementson FCD can also be used to automatically sterilize or(when unremunerated) discourage capital inflows.However, as discussed below, heavy taxationof deposits puts a financial burden on banks andrisks financial disintermediation. Moreover, frequentchanges in reserve requirements are inadvisable be-cause they complicate banks' liquidity management.

Implications for the Payments System

A question also arises whether commercial banksshould be allowed to effect interbank settlement indollars on the books of the central bank, as in Bolivia,Lebanon, Nicaragua, Peru, and Uruguay, or whetherthat should be done on the books of a commercialbank (domiciled locally or overseas; see Table 3).

29For a discussion of these issues see Monetary and ExchangeAffairs Department (1995).

30Thus, in Peru, central bank intervention in the dollar moneymarket is unnecessary, owing to the very high reserve require-ments on FCD, and wide limits on banks' open foreign exchangepositions that facilitate arbitrage between the domestic currencyand foreign currency financial markets.

17


Table 3. Payment and Regulatory Arrangements in Selected Dollarized Economies

Country Currency UsedClearing and Settlement

in Foreign Currency Domestic FCDDomestic Foreign

Currency Loans (FCL)Cross-Border

Deposits (CBD)

Argentina Argentinean peso. Legalcontracts can bedenominated andsettled in U.S. dollars.

Bolivia Boliviano. Legal contractscan be denominated andsettled in U.S. dollars.

Egypt

Lebanon

Egyptian pound.

Lebanese pound.

Domestic wholesaletransactions withdollar-denominatedchecking accounts canbe cleared eitherthrough commercialbank accounts in thecentral bank or usingNew York banksthrough ClearingHouse IntegratedPayment System ordersand settled throughcorrespondentaccounts of bankslocated in New York.

Dollar checks are clearedprivately and settledthrough foreigncurrency accounts inthe Central Bank ofBolivia.

Private banks operate aninformal interbankforeign currencyclearing system.

In 1990, the Bank ofLebanon established aclearing system fordollar-denominatedchecks. Before 1990,check clearing had onlybeen offered by twoprivate institutions.

Developing countries

Restriction on holdings ofFCD were eased in late1978. Foreign currencysavings and timedeposits must bedenominated inconvertible currencies.The use of checkingaccounts denominatedin U.S. dollars is allowedfor domestictransactions.

Restrictions on FCD wereeased in October 1973.From end-1982 to mid-1985, U.S. dollardeposits were notallowed. In 1987, foreigncurrency checkingdeposits werepermitted.

No restrictions.

Even before the civil war,residents were allowedto hold FCD in thedomestic bankingsystem. Dollarizationincreased after the warwhen economicconditions deteriorated.

No restrictions. No restrictions.

No restrictions. No restrictions.

No restrictions.

No restrictions. However,banks may only on-lendup to 65 percent ofFCD

Residents are allowed tohold accounts abroad.

No restrictions.


Peru

Philippines

Uruguay

Nuevo sol. Legal contractscan be denominatedand settled in U.S.dollars.

Philippine peso. Variousdenominations of goldcoins are also legaltender. Since the repealof the UniformCurrency Act of 1996,legal contracts can bedenominated in anycurrency.

Uruguayan peso. Legalcontracts can bedenominated andsettled in U.S. dollars.

Clearing in U.S. dollarstakes place manually atthe clearing housesmanaged by the CentralReserve Bank of Peruwith settlement usingforeign currencyaccounts of commercialbanks at the CRBP. Inaddition, a privateelectronic clearingarrangement is operatedby a group of ten bankswith final settlementusing foreign currencycentral bank accounts.The latter systemaccounts for a smallvolume of totaltransactions.

Dollar-based interbankpayments calledPhilippines ElectronicDollar Transfer System(PEDS) are done throughthe clearing systemoperated, at themoment, by Citibank.The maximum daylightoverdraft that Citibankallows each bank is US$1million. Settlement takesplace at the end of theday, and if anycounterpart fails todeliver on their grossdues, then their due-froms are also withheld,with no payments made.

Clearing in U.S. dollarstakes place manually atthe clearing housesmanaged by the CentralBank of Uruguay withsettlement using foreigncurrency accounts ofcommercial banks at theCentral Bank of Uruguay(Information from1993).

Restrictions on theholdings of FCD wereeased in early 1978.FCD were declaredinconvertible in July 1985and banned in July 1987.While demand and timedeposits were allowedonce more in September1988, savings depositswere allowed in August1990. However, FCDonly regained fullconvertibility in August1990.

Banks obtained largeamounts of foreignexchange from FCDunits after the foreignexchange system wasderegulated in 1992.

Restrictions on holdings ofFCD were eased inOctober 1974. Sincethe 1980s, theUruguayan financialsystem operates as anoffshore center.

Currently, no restrictions.However, from 1979 to1990 (with theexception of the periodin which FCD werebanned), FCD weresubject to marginalreserve requirementsthat were never below90 percent.

No restrictions.

Commercial banks maygrant (1) private sectorloans if serviced usingforeign exchange to besou reed outside thebanking system; (2)short-term loans tofinancial institutions fornormal interbanktransactions; and (3)short-term loans tocommodity and serviceexporters and toimporters. In practice,this regulation does notrestrict FCL

No restrictions.

No restrictions.

No restrictions.(Commercial banks arenot allowed to borrowabroad.)


Table 3 (concluded)

Country Currency UsedClearing and Settlement

in Foreign Currency Domestic FCDDomestic Foreign

Currency Loans (FCL)

Poland Zloty. No domestic clearing inforeign currency.

Vietnam Vietnamese dong. FCD in the domesticbanking system may notbe used for purelydomestic payments.

Economies in transition

Enterprises may maintainforeign currencyaccounts for externalsettlements. SinceDecember 1995 foreignexchange proceeds fromexports can bedeposited in thoseaccounts. Householdsmay maintain FCD butmay not use them toeffect settlementsbetween individuals orfor business activity.Accounts in domesticcurrency are effectivelyconvertible into foreigncurrency by withdrawingzlotys, converting themto dollars, andredepositing them.

Since 1988, householdsand enterprises areallowed to hold FCD indomestic commercialbanks that are licensedto conduct foreignexchange business inVietnam. Organizationsand enterprises mustdeposit all foreignexchange proceeds inthese accounts. Limits onforeign exchangeholdings in such accountsare set by thecommercial banksapplying rules set by thecentral bank. Deposits inexcess of those limitsmust be sold tocommercial banks.

No restrictions.

Cross-BorderDeposits (CBD)

Domestic lending inforeign currency maybe done only for trade-related financing.

Resident households andenterprises whodemonstrate proof ofneed may hold foreignexchange accountsabroad with theNational Bank ofPoland's permission.Balances in theseaccounts may notexceed US$100,000.

Firms in the aviation,shipping, postal, andinsurance sectors, as wellas commercial banks,finance companies, andother firms permitted toopen branches abroadmay be grantedpermission to depositforeign currency receiptsfrom invisibletransactions in foreignaccounts.

Sources: IMF, Exchange Arrangements and Exchange Restrictions Database; and IMF staff.


Prudential Supervision

Neither the central bank nor the commercial bankscan create dollar reserves. Moreover, as illustrated bythe experience of Bolivia, the choice of settlementbank may not have much impact on the pace of dol-larization (Figure 3).31 However, private settlementarrangements may be subject to systemic risk, since ina crisis the clearing bank may be unwilling to extendemergency support, may withdraw its normal settle-ment credit lines, or may fail.32 Moreover, failure tosettle dollar payments could trigger systemic failuresthat would affect both the domestic currency and dol-lar markets. In addition, the central bank losesseignorage if settlement balances are held with an-other bank. In heavily dollarized economies, thoseconsiderations argue in favor of settling interbankdollar balances on the books of the central bank, andin enabling the central bank to provide lender of lastresort services in foreign currency, for which it wouldneed an appropriate cushion of foreign reserves. Fi-nally, the authorities should endeavor to ensure thatpayments services in domestic currency are able tocompete in reliability and efficiency with those in for-eign currency, so as not to provide an additional in-centive for dollarization.33

Figure 3. Bolivia: Dollarization Trends(In percent)

Prudential Supervision

The strategic risks that dollarization can impart tobanks' balance sheets were discussed in Section III.These risks relate primarily to the possibility oflarge capital outflows and exchange rate move-ments. The management of such risks is the broadresponsibility of macroeconomic policy. In addi-tion, however, prudential supervision and regulationneed to take into account the risks of dollarization.The need to enhance banks' capacity to sustain loanlosses after a devaluation adds another argument forbanks in developing countries to exceed the Basleguidelines for capital adequacy.34 Limits on foreign

31The use of dollar sight deposits expanded rapidly in Boliviaafter these accounts were authorized in 1988, notwithstandingthat interbank clearing and settlement of checks drawn on theseaccounts was, until 1991, effected on the basis of private arrange-ments outside the central bank (see Figure 3).

32There is some evidence that Argentine branches of foreignbanks cut their exposure in the interbank market in the aftermathof the Mexican crisis.

33This issue is illustrated by the case of the Philippines, wherepayments dollarization has been encouraged by the fact that theforeign currency payments system—where dollar transactions aresent on an electronic payments system operated by Citibank inManila—technically dominates the peso payments system—which depends on the physical delivery of checks.

34For example, although primarily motivated by the restrictionsimposed by the currency board arrangement rather than by dollar-ization, the capital adequacy ratio was raised to 11.5 percent inArgentina. Moreover, since Argentina applies more stringent cri-

exchange positions—following international stan-dards—should be strictly enforced to contain for-eign exchange risk.35

As noted earlier, reserve requirements on FCD, ifmatched by foreign currency reserves, can limit sys-temic liquidity risk. Since that risk is likely to behigher for FCD than for LCD, from a prudentialpoint of view a case can be made for the former tobear a higher reserve requirement than the latter, asis the case in Bolivia, Honduras, Nicaragua, andPeru (Table 4).36 Thus, the ratio of gross foreign as-sets of the central bank to M3 is substantially higherin Peru than in Bolivia, largely owing to Peru'smuch higher reserve requirements on FCD. Liquid-ity requirements can also perform the role of limit-

teria for weighting risky assets, that ratio is equivalent to about 16percent under the Basle guidelines. In Bolivia, it will gradually beraised to 10 percent, reflecting concerns about dollarization aswell as broader concerns about the soundness of the banking sys-tem. These compare with a Basle guideline of 8 percent.

35These requirements need not exceed those used for nondol-larized economies, since foreign exchange exposure depends on abank's net, rather than gross, position. Moreover, the risk on thebank's gross foreign currency position accrues from credit risk,rather than foreign exchange risk. However, to avoid circumven-tion via derivatives, foreign exchange open positions may need tobe assessed on the basis of risk accounting principles (see Garber,1996).

36However, other considerations—e.g., risk of capital flight—limit the scope for unremunerated reserve requirements onFCD.

21

Source: Central Bank of Bolivia.


V OPERATIONAL ISSUES IN DOLLARIZED ECONOMIES

Table 4. Reserve Requirements on FCD for Selected Countries1(In percent; end-December 1996 unless otherwise noted)

Country

Developing countriesArgentina3

BoliviaEgyptGuinea-BissauHonduras

IndiaJordanLebanonMalawiMalaysia

MaldivesNepalNicaraguaPakistanPeru

PhillippinesSao Tome and PrincipeTanzaniaTurkey

Economies in transitionAlbaniaArmeniaBelarusBulgariaCambodia

Croatia5

Czech RepublicEstoniaLao P.D.R.Latvia

LithuaniaPolandRomaniaRussia6

Slovak Republic

UzbekistanVietnam

OECD countriesGermanyItaly7

JapanUnited Kingdom

ReserveRequirements

on Local CurrencyDeposits (LCD)

17.010.015.025.012.0

10.014.013.020.013.5

35.012.015.05.09.0

17.015.012.08.0

10.012.015.010.05.0

35.911.510.012.08.0

7.017.07.5

10.09.0

25.010.0

2.015.01.30.4

ReserveRequirements

on FCD

17.020.010.025.050.04

0.014.00.0

20.013.5

35.012.025.05.0

45.0

0.030.00.0

11.0

10.012.015.010.05.0

60.011.510.012.08.07.02.0

10.05.09.00.0

10.0

2.015.00.30.4

Currency ofDenomination

of ReserveRequirements

on FCD2

FFFFF

F—FL

FLFLF

—

F—F

FLFLLLF

FLLLL

FLLFLL

F

LLLL

Source: IMF, Monetary and Exchange Affairs Department, Information System on Monetary Instruments.

1Unless otherwise specified, ratio on demand deposits.2L is local currency; F is foreign currency; LF commercial banks can deposit and hold reserves in local or foreign currency.

3Liquidity requirements; reserves on both LCD and FCD may be held abroad.4Of which 12 percentage points must be held in vault cash and deposits in the central bank.

5Weighted average ratio on LCD; reserves on FCD are held abroad.

6As of end-March 1997; ratio on term deposits of more than 90 days.7Applied on new deposits since June 1994.

Ratio of FCD toTotal Deposits

50.092.027.257.026.7

. . .18.553.711.2

...50.37.1

64.4

...74.7

48.445.625.049.3

i i

32.058.031.355.892.7

68.07.0

15.549.052.9

37.617.323.436.511.2

34.729.8

0.62.05.1

15.7

22


Implications for the Central Bank

ing systemic risk—at a lower cost to the banks—ashas been done in Argentina.37

Some highly dollarized countries have imposedrestrictions on FCL (foreign currency loans) to limitcredit risks. In Lebanon, FCL are limited to 60 per-cent of FCD, forcing banks to hold the remainder inforeign currency assets abroad. In Vietnam, FCLmay be given only for trade-related purposes. InMalaysia and the Philippines, dollar loans to the pri-vate sector can be made only to borrowers that gen-erate an income in foreign currency. However, whenFCD are high relative to LCD, strict restrictions onFCL may entail a severe limitation on the availabil-ity of credit for a given volume of total deposits.

Implications for the Central Bank

As lender of last resort, the central bank shouldideally have foreign exchange reserves in sufficientquantity to forestall a run on FCD. The central bank(or the treasury) may also need reserves to supportthe market for dollar-denominated public securi-ties.38 The central bank's ability to secure its role as

37When liquidity requirements are fulfilled with domestic as-sets—rather than foreign liquid assets—the central bank needs tohave sufficient foreign reserves to maintain the liquidity of theseassets in the case of a systemic crisis. Liquidity requirements cantake the form of fixed coefficients that can be differentiated bymaturity of deposits to reflect differences in volatility (as in Ar-gentina) or of limits on maturity mismatches. While more diffi-cult to monitor, the latter also limit interest rate risk. As countriesstrengthen their supervisory capacity and banks improve their in-ternal risk controls, liquidity requirements need to be substitutedby more flexible and comprehensive methods to assess risk. SeeGuide, Nascimento, and Zamalloa (1997).

38Mexico's experience with tesobonos in 1994 suggests thatgovernment bonds denominated in foreign currency may be sub-

lender of last resort, or to ensure orderly conditionsin the market for (dollar-denominated) public sectordebt, can be enhanced by access to emergency creditlines from other central banks or from commercialbanks.39 In addition, allowing sound foreign banks,which can receive head office support in an emer-gency, to operate can also limit the need for centralbank systemic support.

Supervisory authorities need to be particularlyvigilant to ensure that financial institutions have thefinancial skills and the risk management arrange-ments to manage effectively the risks inherent in op-erating in a dual currency system. In particular, for-eign currency exposures need to be monitoredclosely. These considerations apply also to bankingsystems that operate heavily in foreign currencieseven if the economies in which they function are notdollarized. However, the problem is likely to bemore widespread in dollarized economies, wherebanks are less likely to have a choice as regards thecurrency denomination of their operations.

Since a run on foreign currency bank deposits candestabilize a banking system (including the localcurrency segment), there is a case for including FCDin a country's deposit insurance arrangements. Careshould be taken, however, to price that insuranceproperly, in order to avoid cross-subsidization.

ject to speculative attacks, particularly when they have short ma-turities and inadequate foreign reserve backing.

39Thus, in the wake of the Mexican crisis in 1994, a number ofAsian central banks established a protocol for mutual liquiditysupport. Argentina established an arrangement that facilitatesemergency borrowing from foreign commercial banks by localcommercial banks. Although these examples are not specific todollarized economies, similar arrangements can be tailored tothem. See Balino and others (1997).

23


VI Measures to Affect Dollarization

The prerequisite to containing dollarization in aneconomy is always a sound macroeconomicpolicy. Also, institutional arrangements that bolsterconfidence in the maintenance of price stability—such as an independent central bank with a clearprice stability mandate—can enhance confidence inthe domestic currency and thereby reduce the degreeof dollarization. As is argued above, however, it ispossible for an economy to remain dollarized evenafter stabilization. This section discusses the variousalternative measures available to limit dollarizationand considers their effectiveness and drawbacks. Itconcludes that the most effective "dedollarization"measures are likely to be those that limit the need fordollar instruments. The least effective are those thatdirectly restrict dollarization.

Alternative Financial InstrumentsTo limit dollarization, authorities may promote al-

ternative financial instruments, such as stocks, mu-tual fund shares, corporate finance bonds, and asset-backed securities and also facilitate the use ofspecific indexed instruments or simple derivatives(swaps, futures, and options). These instruments,which are likely to be used mostly by enterprises andlarge investors, unbundle risk coverage from thetransaction and store-of-value functions of currency.By limiting the replacement of local currency by for-eign currency, including reserve money, those instru-ments in principle can enhance the scope for mone-tary policy. However, they are likely to be onlypartial substitutes for FCD and FCL—owing, for in-stance, to high transaction costs—and may be diffi-cult to develop in unsophisticated markets. Further-more, derivatives entail risks, to which the authoritiesshould ensure that banks are not unduly exposed.

In several countries, dollar-indexed financial in-struments provide an alternative to FCD. For exam-ple, in Bolivia and Nicaragua, dollar-indexed de-posits coexist with dollar deposits.40 The supply of

Figure 4. Chile: Indexationand Dollarization(In billions of U.S. dollars)

the former may be easier to influence because theyare settled in domestic currency, which is under thecontrol of the central bank. However, because of theuncertainty associated with the exchange rate atwhich dollar-indexed instruments can be effectivelyconverted into dollars, particularly under systemiccrisis conditions, these instruments are imperfectsubstitutes for FCD.41

Indexation with alternative indices, such as theprice level or a short-term interest rate, can also re-duce the risk associated with real returns on domesticcurrency deposits. In Brazil, while FCD have been

40The "mexdollar" system, which was widely used in Mexicountil the forced conversion of deposits in August 1982, and the"Patzam" in Israel provide other illustrations.

41Thus, in Bolivia, their demand is mostly limited to public en-terprises, which are required by law to hold their deposits in localcurrency. Similarly, in Nicaragua, dollar-indexed deposits pay apremium over dollar deposits.

24

Sources: IMF, International Financial Statistics (IFS); and nationalauthorities.


strictly restricted, the relatively moderate capitalflight suggests that indexation adequately met thepublic's demand for inflation protection.42 In Chile, ashas happened in some highly dollarized economies,price indexation is still widely used, despite the coun-try's success in bringing inflation under control.43 Al-though FCD are also available, they are small (Figure4), which suggests that price indexation can be an ef-fective substitute for dollarization. However, in con-sidering indexation as an alternative to dollarization,the authorities need to evaluate carefully the benefitsand risks. The latter include the danger of indexationbeing extended to the labor market and the difficultiesthis might create, including for bringing inflationdown.

Policies That Create an InterestRate Wedge

To the extent that portfolio choices are sensitiveto relative prices, the dollarization of deposits can,in principle, be reduced through policies that createa wedge in favor of local currency interest rates.However, when those policies take the form of atax on foreign currency intermediation, they runthe risk of driving deposits overseas and reducingdomestic lending.44 Interest rate wedges will be themore effective in reversing dollarization to the ex-tent that FCD are closer substitutes for LCD thanfor CBD, which can be the case in economieswhere confidence in the domestic economy hasbeen restored.

Financial liberalization can permanently changethe level of interest rates. Reflecting this, Egypt ex-perienced a dramatic and lasting reversal in dollar-ization after liberalizing interest rates in 1991.45

FCD decreased as a percentage of total depositsfrom over 60 percent in 1991 to less than 30 percentin 1996 (Figure 5).

Figure 5. Egypt: Asset Substitutionand Real Deposit Rates1(In percent)

42In Brazil, both price indexation and interest rate indexationwere broadly used. In particular, the indexation of deposits tothe overnight interest rate protected the purchasing power ofLCD throughout the turbulent inflationary period of the 1980s.However, the need to limit risks incurred by financial interme-diaries in the overnight market eventually undermined the mon-etary anchor, since it induced pervasive monetary accommoda-tion and led to the de facto creation of indexed money. SeeGarcia (1996).

43Price indexation has been facilitated by the introduction in1967 of a unit of account, the UF, that is published by the centralbank daily on the basis of the consumer price index.

44Also, since "dedollarization" measures are very difficult toapply to cash, part of the flow out of FCD can move to DCC.

45Other countries where financial liberalization may have con-tributed to limit dollarization include Armenia, Estonia, Lithua-nia, and Poland (see Sahay and Vegh, 1996).

Monetary policy may also induce interest ratewedges, although these are likely to be temporary.46

Thus, in Hungary, the sharp increase in the interestrate spread during 1995-96 became ultimately un-sustainable because it was associated with large cap-ital inflows that could not be fully sterilized, and thewedge had only a small and transitory impact ondollarization (Figure 6).

Modifying the currency composition of domesticpublic debt or introducing differential reserve re-quirements can introduce a permanent wedge. In-creasing the share of local currency domestic publicdebt at the expense of foreign currency domesticdebt can increase interest rates in local currency andlower those in foreign currency. However, the higherinterest rate on local currency loans reduces at thesame time the demand for such loans and increasesthe share of FCL. In addition, the interest costs asso-ciated with such operations can quickly escalate.

Setting differential remuneration rates on reserverequirements on FCD introduces a wedge in banks'intermediation spreads, thereby in principle affect-ing dollarization measured both in terms of deposits

46 A too tight monetary stance, if sustained, can have other ad-verse macroeconomic and financial implications, including an in-crease in the burden of public domestic debt.

25

Sources: Central Bank of Egypt; and IMF, IFS.1Real deposit rate is measured as the nominal term deposit

rate adjusted by inflation in the previous 12 months.


VI MEASURES TO AFFECT DOLLARIZATION

Sources: IMF, IFS; and National Bank of Hungary.1Interest rate differential measured as domestic currency

deposit rates as a percentage of Hungarian deposit rates adjust-ed for the previous 12-month depreciation rate. The use of theforward 12-month depreciation rate to calculate the interest dif-ferential does not yield a higher correlation coefficient (inabsolute terms) between the interest rate differential and theratio of FCD to total deposits.

and loans.47 Thus, using such a technique Israel suc-ceeded in the late 1980s in encouraging the substitu-tion of dollar-indexed deposits ("Patzam") for dollardeposits ("Patam").

Documents

OCCASIONAL PAPER · 2020. 4. 1. · NIR Net international reserves RR Reserve requirements V ©International Monetary Fund. Not for Redistribution. This page intentionally left blank